Intelligent analysis method and system based on industrial data visualization

Abstract

The present invention relates to the field of data processing technology, and in particular to an intelligent analysis method and system based on industrial data visualization. The method includes: preprocessing a data set to output a reference data set; using the reference data set to train an analysis model to be trained to output an analysis model; respectively obtaining the output results of the two same inputs of the analysis model before and after the model is updated; determining the first corresponding backup frequency of the analysis model update data based on the difference in the byte volume of the output results of the analysis model and the number of the same error types in the analysis model; obtaining the response delay duration of the analysis model after running a single cycle according to the first corresponding backup frequency of the analysis model update data; determining the second corresponding backup frequency of the analysis model update data based on the response delay duration of the analysis model; the present invention achieves an improvement in the running stability of the analysis model.

Description

Technical Field

[0001] The present invention relates to the technical field of visualized intelligent analysis, and in particular to an intelligent analysis method and system based on industrial data visualization. Background Art

[0002] With the rapid development of information technology, intelligent manufacturing has become an important direction for industrial upgrading. Intelligent analysis methods and systems based on industrial data visualization are of great significance to the development of intelligent manufacturing. Through the combination of data visualization and intelligent analysis, enterprises can better understand the real-time status and problems of the production line, formulate optimization and improvement measures, and improve production efficiency and quality.

[0003] Chinese Patent Publication No.: CN114925835A discloses a method and system for intelligently generating an industrial map. Through industrial source data acquisition, data preprocessing, and intelligent analysis and mining, it constructs and generates a three-dimensional integrated industrial map of industrial policy maps, industrial chain maps, and industrial development history maps. Through this system, practitioners who are not in this field can quickly and easily generate industrial maps under corresponding industrial categories according to the needs of industrial maps, greatly improving the efficiency of industrial data analysis. The system will automatically merge and organize the map data, display it visually, and export it in a variety of file formats for enterprises to view, providing a good user experience. It can be seen that the method and system for intelligently generating industrial maps have the problem that the output data of data training is updated to the artificial intelligence question-and-answer model, and the updated data conflicts with the data of the artificial intelligence question-and-answer model itself, causing errors in the system operation, thereby reducing the stability of the system operation. Summary of the invention

[0004] To this end, the present invention provides an intelligent analysis method based on industrial data visualization, so as to overcome the problem in the prior art that the output data of data training is updated into the artificial intelligence question-answering model, and the updated data conflicts with the data of the artificial intelligence question-answering itself, causing errors in system operation, thereby resulting in a decrease in the stability of system operation.

[0005] To achieve the above-mentioned purpose, the present invention provides an intelligent analysis method based on industrial data visualization, comprising the following steps: obtaining a number of basic industrial data to form a data set, and preprocessing the data set to output a reference data set; performing feature analysis on the reference data set to determine the analysis model to be trained, and using the reference data set to train the analysis model to output the analysis model; performing statistics on the running errors of the analysis model to update the analysis model; determining the corresponding number of repeated trainings of the data training process based on the average byte difference of the output results of the analysis model, or respectively obtaining the output results of two identical inputs of the analysis model before and after the model update; determining the update mode of the analysis model based on the difference in the byte amount of the output results of the analysis model, or determining the first corresponding backup frequency of the analysis model update data based on the difference in the byte amount of the output results of the analysis model and the number of the same error types occurring in the analysis model; obtaining the response delay duration of the analysis model after running a single cycle according to the first corresponding backup frequency of the analysis model update data; determining the second corresponding backup frequency of the analysis model update data based on the response delay duration of the analysis model; wherein the first corresponding backup frequency of the analysis model update data is greater than the second corresponding backup frequency of the analysis model update data.

[0006] Furthermore, the step of determining the number of repeated training times of the data training process includes:

[0007] Obtaining the byte amount of the output result of the analysis model under the same input amount condition to calculate the average byte difference amount of the output result of the analysis model;

[0008] Comparing the average byte difference of the output result of the analysis model with a preset first average difference and a preset second average difference respectively;

[0009] If the average byte difference of the output result of the analysis model is greater than the preset second average difference, the analysis model is trained according to the number of repeated training times corresponding to the preset second average difference;

[0010] Wherein, the preset first average difference is smaller than the preset second average difference.

[0011] Furthermore, the corresponding number of repeated training times is determined by the difference between the average byte difference of the output result of the analysis model and the preset second average difference.

[0012] Furthermore, the step of determining the method of updating the model includes:

[0013] If the average byte difference of the output result of the analysis model is greater than the preset first average difference and less than or equal to the preset second average difference, it is preliminarily determined that the accuracy of the analysis model update does not meet the requirements;

[0014] Obtaining output results of two identical input quantities of the analysis model before and after the update, respectively, and calculating the difference in byte quantities of the output results of the analysis model;

[0015] Comparing the difference in the byte amount of the output result of the analysis model with a preset first difference amount and a preset second difference amount respectively;

[0016] If the difference in the byte amount of the output result of the analysis model is greater than the preset second difference, it is secondarily determined that the accuracy of the analysis model update does not meet the requirements, and the analysis model is updated according to the update method of the analysis model.

[0017] Further, the updating method of the analysis model is: when the output result of the analysis model is a data set, the maximum number of update types of the analysis model is configured as a first update number; when the output result of the analysis model is a single data, the maximum number of update types of the analysis model is configured as a second update number;

[0018] The first update quantity is greater than the second update quantity.

[0019] Further, the step of determining the first corresponding backup frequency includes:

[0020] If the difference in bytes of the output results of the analysis model is greater than the preset first difference and less than or equal to the preset second difference, it is preliminarily determined that the comprehensiveness of the regression test does not meet the requirements;

[0021] Obtain the number of the same error types in the analysis model for calculation;

[0022] Comparing the number of the same error type occurring in the analysis model with a preset number;

[0023] If the number of the same error type appearing in the analysis model is greater than the preset number, it is determined that the comprehensiveness of the regression test does not meet the requirements, and the first corresponding backup frequency is determined.

[0024] Furthermore, the first corresponding backup frequency is determined by the difference between the number of the same error type occurring in the analysis model and the preset number.

[0025] Further, the step of determining the second corresponding backup frequency includes:

[0026] Comparing the response delay time of the analysis model with a preset delay time;

[0027] If the response delay time of the analysis model is greater than the preset delay time, it is determined that the real-time performance of the data update does not meet the requirement, and the second corresponding backup frequency is determined.

[0028] Furthermore, the second corresponding backup frequency is determined by a difference between a response delay duration of the analysis model and the preset delay duration.

[0029] The present invention also provides an intelligent analysis system based on an intelligent analysis method of industrial data visualization, comprising:

[0030] A data collection module is used to collect some basic industry data to form a data set;

[0031] A data preprocessing module, connected to the data acquisition module, for preprocessing the data set to output a control data set;

[0032] A model training module, which is connected to the data acquisition module and the data preprocessing module respectively, and is used to perform feature analysis on the industry feature data in the reference data set to determine the analysis model to be trained, and to train the analysis model to be trained using the reference data set to output an analysis model;

[0033] A storage module, which is connected to the data acquisition module, the data preprocessing module and the model training module, respectively, and is used to store the training information and update information of the data set, the control data set and the analysis model, respectively;

[0034] A control module, which is respectively connected to the data acquisition module, the data preprocessing module, the storage module and the model training module, and is used to determine the number of repeated training times of the data training process according to the average byte difference of the output results of the analysis model, or to determine the model update method according to the average byte difference of the output results of the analysis model and the difference in the byte amount of the output results of the analysis model, and to determine the first corresponding backup frequency of the analysis model update data according to the difference in the byte amount of the output results of the analysis model and the number of the same error types occurring in the analysis model, and to determine the second corresponding backup frequency of the analysis model update data according to the response delay duration of the analysis model.

[0035] Compared with the prior art, the beneficial effect of the present invention lies in that the analysis method of the present invention determines the number of repeated training times of the data training process according to the byte average difference of the output result of the analysis model by setting a preset first average difference amount and a preset second average difference amount, thereby reducing the impact of errors in system operation caused by conflicts between the output data of data training and the data of the artificial intelligence question and answer model, thereby reducing the stability of system operation; by setting the preset first difference amount and the preset second difference amount and determining the model update method, the impact of reduced accuracy of the analysis model update caused by excessive changes in output data when the output data is a single data due to excessive update range is reduced, thereby improving the stability of the analysis model operation.

[0036] Furthermore, by setting a preset first average difference and a preset second average difference, the number of repeated training times of the data training process is determined according to the byte average difference of the output result of the analysis model, thereby reducing the impact of errors in system operation caused by conflicts between the output data of the data training and the data of the artificial intelligence question and answer model, thereby reducing the stability of the system operation. This further improves the stability of the analysis model operation.

[0037] Furthermore, the analysis method of the present invention sets a preset first difference amount and a preset second difference amount, and determines the model update method, thereby reducing the impact of excessive changes in output data when the output data is a single data due to an excessively large update range, which leads to a decrease in the accuracy of the analysis model update, and further improves the operating stability of the analysis model.

[0038] Furthermore, the analysis method described in the present invention sets a preset byte amount. If the updated related data is not backed up in time when the analysis model is updated, the comprehensiveness of the analysis model test during regression testing will be insufficient, which will cause omissions of certain functional tests and cause the analysis model to encounter unknown errors in actual operation. As a result, the accuracy of the next model update of the analysis model will not meet the requirements. By increasing the backup frequency of the analysis model update data, the operation stability of the analysis model is further improved.

[0039] Furthermore, the analysis method of the present invention sets a preset delay duration and determines the second corresponding backup frequency of the analysis model update data according to the response delay duration of the analysis model. This reduces the impact of excessive backup frequency of the analysis model update data, excessive system data processing, and increased system response time, thereby reducing the real-time performance of data updates, thereby further improving the operational stability of the analysis model. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 This is an overall flow chart of an intelligent analysis method based on industrial data visualization according to an embodiment of the present invention;

[0041] Figure 2 This is an overall structural block diagram of an intelligent analysis system based on industrial data visualization according to an embodiment of the present invention;

[0042] Figure 3 A flowchart of determining the number of repeated training times of a data training process according to an embodiment of the present invention;

[0043] Figure 4 The present invention is a flowchart of a method for determining model update according to an embodiment of the present invention. DETAILED DESCRIPTION

[0044] In order to make the objects and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0045] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the protection scope of the present invention.

[0046] In addition, it should be noted that in the description of the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For those skilled in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0047] It should be pointed out that the data in this embodiment are obtained by the intelligent analysis method based on industrial data visualization of the present invention according to the historical test data and the data statistics, testing and comprehensive analysis based on the test results; the intelligent analysis method based on industrial data visualization of this embodiment is based on the statistics, testing and calculation of the average byte difference of the output results of the analysis model of 1742 cases accumulated within 90 days, the difference in the byte amount of the output results of the analysis model, the number of the same error types in the analysis model and the response delay time of the analysis model before the current use, and comprehensively determines the values ​​of various preset parameter standards for the intelligent analysis method based on industrial data visualization. It can be understood by those skilled in the art that the intelligent analysis method and system based on industrial data visualization described in this embodiment for determining a single parameter is to select the value with the highest number of times as the preset standard parameter according to the data distribution, and the parameter determination method only needs to meet the different specific situations in the single determination process that can be clearly defined by the obtained value.

[0048] See also Figure 1 , Figure 2 , Figure 3 as well as Figure 4 As shown, they are respectively an overall flow chart of an intelligent analysis method based on industrial data visualization according to an embodiment of the present invention, an overall structural block diagram of an intelligent analysis system based on industrial data visualization according to an embodiment of the present invention, a flow chart for determining the number of repeated training times of a data training process, and a flow chart for determining a method for updating a model. An embodiment of the present invention provides an intelligent analysis method based on industrial data visualization, comprising the following steps:

[0049] Step S1, obtaining a number of basic industry data to form a data set, and preprocessing the data set to output a reference data set;

[0050] Step S2, performing feature analysis on the reference data set to determine an analysis model to be trained, and using the reference data set to train the analysis model to be trained to output an analysis model;

[0051] Step S3, collecting statistics on the running errors of the analysis model to update the analysis model;

[0052] Step S4, determining the corresponding number of repeated training times of the data training process based on the average byte difference of the output results of the analysis model, or respectively obtaining the output results of two identical inputs of the analysis model before and after the model is updated;

[0053] Step S5, determining the update method of the analysis model based on the difference in the byte volume of the output result of the analysis model, or determining the first corresponding backup frequency of the analysis model update data based on the difference in the byte volume of the output result of the analysis model and the number of the same error type occurring in the analysis model;

[0054] Step S6, obtaining a response delay duration of the analysis model after running a single cycle according to the first corresponding backup frequency of the analysis model update data;

[0055] Step S7, determining a second corresponding backup frequency of the analysis model update data based on the response delay duration of the analysis model;

[0056] The first corresponding backup frequency of the analysis model update data is greater than the second corresponding backup frequency of the analysis model update data.

[0057] Specifically, the model update data means the relevant data of the analysis model during the model update process. The model update data includes the operation abnormality data of the analysis model, the model training data, the parameters of the analysis model and the regression test performance data.

[0058] Specifically, the calculation formula for the average byte difference of the output results of the analysis model is:

[0059]

[0060] Where Q is the average byte difference of the output results of the analysis model, |A x+1 -A x |The absolute value of the difference between the byte size of the output result of the x+1th analysis model and the byte size of the output result of the xth analysis model under the same input volume, where L is the number of outputs of the analysis model.

[0061] Specifically, the calculation formula for the difference in the byte volume of the output results of the analysis model is:

[0062] M=|B1-B2|

[0063] Among them, M is the difference in the byte size of the output results of the analysis model, kb1 is the byte size of the output results of the analysis model after the update, and kb2 is the byte size of the output results of the analysis model for the same input before and after the update.

[0064] Specifically, the number of the same error types occurring in the analysis model means the number of errors occurring in the current model running process that are the same as the number of errors occurring in the previous model running process.

[0065] Specifically, the types of errors that occur in analytical models include unclear generated bar charts or graphs, missing data in output results, and changes in the operating efficiency of the analytical model.

[0066] Specifically, the calculation formula for the response delay time of the analysis model is:

[0067] W=D1-D2

[0068] Wherein, W is the response delay time of the analysis model, D1 is the actual response time of the analysis model, and D2 is the standard response time of the analysis model.

[0069] Specifically, basic industry data include investment and financing data, policy data, scientific research project data, and patent data under several industry categories.

[0070] Specifically, the basic industry data may be in the form of structured data, semi-structured data, and unstructured data.

[0071] Specifically, structured data refers to data with a fixed format and structure, such as a table or a CSC file.

[0072] Specifically, the semi-structured data may be in XML format or JSO format.

[0073] Specifically, unstructured data can be text, images, and audio.

[0074] Specifically, the updating process of the analysis model is to statistically analyze the running errors of the analysis model for several times, extract features from the running errors, and determine the fit of the analysis model. When the fit of the analysis model is lower than 0.5, the parameters of the analysis model are updated.

[0075] Specifically, the parameters of the analysis model may be the algorithm of the analysis model, the number of mappings, and the number of iterations.

[0076] Specifically, the process of preprocessing the data set includes operations such as denoising, filling missing values, and standardization.

[0077] Specifically, the essence of performing feature analysis on the reference data set to determine the analysis model to be trained is to select an existing basic analysis model stored in the system according to the data characteristics of the reference data set.

[0078] Specifically, the specific principle of performing feature analysis on the reference data set to determine the analysis model to be trained is to extract meaningful information from the original data in the reference data set and convert it into features that can be understood and used by the existing model. For those skilled in the art, the process of determining the analysis model to be trained is a conventional technical means well known to those skilled in the art, so the specific process will not be repeated here.

[0079] The analysis method described in the present invention sets a preset first average difference amount and a preset second average difference amount, and determines the number of repeated training times of the data training process according to the byte average difference amount of the output result of the analysis model, thereby reducing the impact of errors in system operation caused by conflicts between the updated data and the data of the artificial intelligence question and answer model when the output data of the data training is updated to the artificial intelligence question and answer model, thereby reducing the impact of the stability of the system operation being reduced due to errors in system operation; by setting the preset first difference amount and the preset second difference amount, and determining the model update method, the impact of reduced accuracy of the analysis model update caused by excessive changes in output data when the output data is a single data due to excessive update range is reduced, thereby improving the stability of the analysis model operation.

[0080] See also Figure 3 As shown, the step of determining the number of repeated training times of the data training process includes:

[0081] Obtaining the byte amount of the output result of the analysis model under the same input amount condition to calculate the average byte difference amount of the output result of the analysis model;

[0082] Comparing the average byte difference of the output result of the analysis model with a preset first average difference and a preset second average difference respectively;

[0083] If the average byte difference of the output result of the analysis model is greater than the preset second average difference, training the analysis model according to the corresponding number of repeated training times;

[0084] Wherein, the preset first average difference is smaller than the preset second average difference.

[0085] See also Figure 3 As shown, the corresponding number of repeated training times is determined by the difference between the average byte difference of the output result of the analysis model and the preset second average difference.

[0086] Specifically, when the average byte difference of the output result of the analysis model is greater than the preset second byte difference, it is determined that the stability of the system operation does not meet the requirements.

[0087] Specifically, if the average byte difference of the output result of the analysis model is greater than the preset second average difference, the specific process of determining the corresponding number of repeated training times of the data training process is:

[0088] Specifically, if △Q≤△Q0, the number of repeated training times of the data training process is adjusted to the first number using the preset first number adjustment coefficient;

[0089] If △Q>△Q0, the control module uses the preset second number adjustment coefficient to adjust the number of repeated training times of the data training process to the second number;

[0090] Specifically, the corresponding number of repeated training times of the data training process includes a first number and a second number.

[0091] Specifically, the average byte difference of the output results of the analysis model is recorded as Q, the preset first average difference is recorded as Q1, and the preset second average difference is recorded as Q2, Q1=15kb, Q2=20kb, the difference between the average byte difference of the output results of the analysis model and the preset second average difference is recorded as △Q, △Q=Q-Q2 is set, △Q0 is the preset average difference difference, △Q0=5kb, the preset first number adjustment coefficient is recorded as α1, and the preset second number adjustment coefficient is recorded as α2, wherein 1<α1<α2, α1=1.1, α2=1.2, the number of repeated training times of the data training process is recorded as E, and the corresponding number of repeated training times of the data training process is recorded as E'. Set E'=E×αi, wherein αi is the preset i-th adjustment coefficient, and set i=1, 2.

[0092] By setting a preset first average difference and a preset second average difference, the number of repeated training times of the data training process is determined according to the byte average difference of the output results of the analysis model, thereby reducing the impact of errors in system operation caused by conflicts between the output data of data training and the data of the artificial intelligence question and answer model, thereby reducing the stability of the system operation. This further improves the stability of the analysis model operation.

[0093] Please continue reading Figure 4 As shown, the determination step of determining the method of updating the model includes:

[0094] If the average byte difference of the output result of the analysis model is greater than the preset first average difference and less than or equal to the preset second average difference, it is preliminarily determined that the accuracy of the analysis model update does not meet the requirements;

[0095] Obtaining output results of two identical input quantities of the analysis model before and after the update, respectively, and calculating the difference in byte quantities of the output results of the analysis model;

[0096] Comparing the difference in the byte amount of the output result of the analysis model with a preset first difference amount and a preset second difference amount respectively;

[0097] If the difference in the byte amount of the output result of the analysis model is greater than the preset second difference, it is secondarily determined that the accuracy of the analysis model update does not meet the requirements, and the analysis model is updated according to the update method of the analysis model.

[0098] Please continue reading Figure 4 As shown, the update method of the analysis model is: when the output result of the analysis model is a data set, the maximum number of update types of the analysis model is configured as the first update number; when the output result of the analysis model is a single data, the maximum number of update types of the analysis model is configured as the second update number;

[0099] The first update quantity is greater than the second update quantity.

[0100] Specifically, the difference in bytes of the output results of the analysis model is recorded as M, the preset first difference is recorded as M1, and the preset second difference is recorded as M2, wherein M1<M2, and M1=1kb, M2=3kb.

[0101] Specifically, the first update quantity is 8 and the second update quantity is 4.

[0102] Specifically, the update types of the analysis model include the mapping relationship of the analysis model, the operation performance parameters of the analysis model, the range of the input quantity of the analysis model, and the type of the output result of the analysis model.

[0103] The analysis method of the present invention sets a preset byte amount and determines the first corresponding backup frequency of the analysis model update data according to the amount of lost bytes of the data after the model is updated. This reduces the impact of insufficient regression testing during system updates, which leads to missing tests of some functions, causing the system to encounter unknown errors in actual operation, resulting in data loss and reduced comprehensiveness of regression testing, thereby further improving the operational stability of the analysis model.

[0104] Please continue reading Figure 1 As shown, the step of determining the first corresponding backup frequency includes:

[0105] If the difference in bytes of the output results of the analysis model is greater than the preset first difference and less than or equal to the preset second difference, it is preliminarily determined that the comprehensiveness of the regression test does not meet the requirements;

[0106] Obtain the number of the same error types in the analysis model for calculation;

[0107] Comparing the number of the same error type occurring in the analysis model with a preset number;

[0108] If the number of the same error type appearing in the analysis model is greater than the preset number, it is determined that the comprehensiveness of the regression test does not meet the requirements, and the first corresponding backup frequency is determined.

[0109] Please continue reading Figure 1As shown, the first corresponding backup frequency is determined by the difference between the number of the same error type occurring in the analysis model and the preset number.

[0110] Specifically, the specific process of determining the first corresponding backup frequency of the analysis model update data is as follows:

[0111] If △R≤△R0, the standard backup frequency of the analysis model update data is adjusted to the first frequency using the preset first backup frequency adjustment coefficient;

[0112] If △R>△R0, the standard backup frequency of the analysis model update data is adjusted to the second frequency using the preset second backup frequency adjustment coefficient;

[0113] Specifically, the first corresponding backup frequency of the analysis model update data includes a first frequency and a second frequency.

[0114] Specifically, the number of the same error type appearing in the analysis model is denoted as R, the preset number is denoted as R0, R0=4, the difference between the number of the same error type appearing in the analysis model and the preset number is denoted as △R0, △R=R-R0 is set, △R0 is the preset number difference, △R0=2, the preset first backup frequency adjustment coefficient is denoted as γ1, the preset second backup frequency adjustment coefficient is denoted as γ2, wherein 1<γ1<γ2, γ1=1.15, γ2=1.3, the standard backup frequency of the analysis model update data is denoted as T, the first corresponding backup frequency of the analysis model update data is denoted as T', T'=T×γp is set, wherein γp is the preset pth backup frequency adjustment coefficient, and p is set as 1, 2.

[0115] The analysis method described in the present invention sets a preset byte amount. If the updated related data is not backed up in time when the analysis model is updated, the comprehensiveness of the analysis model test during regression testing will be insufficient, which will cause omissions of certain functional tests and cause the analysis model to encounter unknown errors in actual operation. As a result, the accuracy of the next model update of the analysis model will not meet the requirements. By increasing the backup frequency of the analysis model update data, the operation stability of the analysis model is further improved.

[0116] Please continue reading Figure 1 As shown, the step of determining the second corresponding backup frequency includes:

[0117] Comparing the response delay time of the analysis model with a preset delay time;

[0118] If the response delay time of the analysis model is greater than the preset delay time, it is determined that the real-time performance of the data update does not meet the requirement, and the second corresponding backup frequency is determined.

[0119] Please continue reading Figure 1 As shown, the second corresponding backup frequency is determined by the difference between the response delay time of the analysis model and the preset delay time.

[0120] Specifically, the specific process of determining the second corresponding backup frequency of the analysis model update data is as follows:

[0121] Specifically, if ΔW≤ΔW0, the standard backup frequency of the analysis model update data is adjusted to the third frequency using the preset fourth backup frequency adjustment coefficient;

[0122] If ΔW>ΔW0, the standard backup frequency of the analysis model update data is adjusted to the fourth frequency using the preset third backup frequency adjustment coefficient.

[0123] Specifically, the second corresponding backup frequency of the analysis model update data includes a third frequency and a fourth frequency.

[0124] Specifically, the response delay time of the analysis model is denoted as W, the preset delay time is denoted as W0, and W0 is set to 0.1s. The difference between the response delay time of the analysis model and the preset delay time is denoted as △W, and △W is set to W-W0. △W0 is the preset delay time difference, △W0=0.05s, the preset third backup frequency adjustment coefficient is denoted as γ3, and the preset fourth backup frequency adjustment coefficient is denoted as γ4, wherein 0<γ3<γ4<1, the second corresponding backup frequency T" of the analysis model update data is set to T"=T'×γj, wherein γj is the preset jth backup frequency adjustment coefficient, and j is set to 3, 4.

[0125] The analysis method of the present invention sets a preset delay time and determines the second corresponding backup frequency of the analysis model update data according to the response delay time of the analysis model. This reduces the impact of excessive backup frequency of the analysis model update data, excessive system data processing, and increased system response time, thereby reducing the real-time performance of data updates, thereby further improving the operating stability of the analysis model.

[0126] The embodiment of the present invention further provides an intelligent analysis system for implementing the above-mentioned intelligent analysis method based on industrial data visualization, comprising:

[0127] A data collection module is used to collect some basic industry data to form a data set;

[0128] A data preprocessing module, connected to the data acquisition module, for preprocessing the data set to output a control data set;

[0129] A model training module, which is connected to the data acquisition module and the data preprocessing module respectively, and is used to perform feature analysis on the industry feature data in the reference data set to determine the analysis model to be trained, and to train the analysis model to be trained using the reference data set to output an analysis model;

[0130] A storage module, which is connected to the data acquisition module, the data preprocessing module and the model training module, respectively, and is used to store the training information and update information of the data set, the control data set and the analysis model, respectively;

[0131] A control module, which is respectively connected to the data acquisition module, the data preprocessing module, the storage module and the model training module, and is used to determine the number of repeated training times of the data training process according to the average byte difference of the output results of the analysis model, or to determine the model update method according to the average byte difference of the output results of the analysis model and the difference in the byte amount of the output results of the analysis model, and to determine the first corresponding backup frequency of the analysis model update data according to the difference in the byte amount of the output results of the analysis model and the number of the same error types occurring in the analysis model, and to determine the second corresponding backup frequency of the analysis model update data according to the response delay duration of the analysis model.

[0132] Specifically, the intelligent analysis system also includes a question-and-answer module that is respectively connected to the model training module, the storage module, and the control module to provide AI question-and-answer functions.

[0133] Example 1

[0134] In the present embodiment 1, when R>R0, the standard backup frequency T of the analysis model update data is adjusted according to the difference △R0 between the number of the same error type appearing in the analysis model and the preset number. The number of the same error type appearing in the analysis model is recorded as R, the preset number is recorded as R0, and the difference between the number of the same error type appearing in the analysis model and the preset number is recorded as △R0. Set △R=R-R0, △R0 is the preset number difference, △ preset first backup frequency adjustment coefficient is recorded as γ1, and preset second backup frequency adjustment coefficient is recorded as γ2, wherein 1<γ1<γ2, the standard backup frequency of the analysis model update data is recorded as T, and the first corresponding backup frequency of the analysis model update data is recorded as T', T=30 times / h, and R=8 in the present embodiment 1,

[0135] In this embodiment 1, ΔR=8-4=4 is obtained, and ΔR>ΔR0 is determined and the standard backup frequency of the analysis model update data is adjusted to the second frequency T' using the preset second backup frequency adjustment coefficient γ2, and T'=30 times / h×1.3=39 times / h is calculated.

[0136] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

Claims

1. An intelligent analysis method based on industrial data visualization, characterized in that: The following steps are involved: Acquiring a number of basic industry data to form a data set, and preprocessing the data set to output a control data set; Performing feature analysis on the reference data set to determine an analysis model to be trained, and using the reference data set to train the analysis model to be trained to output an analysis model; Collecting statistics on the running errors of the analysis model to update the analysis model; Determine the corresponding number of repeated trainings of the data training process based on the average byte difference of the output results of the analysis model; updating the analysis model based on the difference in the byte amount of the output result of the analysis model; Determine the first corresponding backup frequency of the analysis model update data based on the difference in the byte volume of the output result of the analysis model and the number of the same error types occurring in the analysis model, and obtain the response delay duration of the analysis model after running a single cycle according to the first corresponding backup frequency of the analysis model update data; Determining a second corresponding backup frequency of the analysis model update data based on the response delay duration of the analysis model; wherein the first corresponding backup frequency of the analysis model update data is greater than the second corresponding backup frequency of the analysis model update data; The step of determining the number of repeated training times of the data training process comprises: Obtaining the byte amount of the output result of the analysis model under the same input amount condition to calculate the average byte difference amount of the output result of the analysis model; Comparing the average byte difference of the output result of the analysis model with a preset first average difference and a preset second average difference respectively; If the average byte difference of the output result of the analysis model is greater than the preset second average difference, training the analysis model according to the corresponding number of repeated training times; Wherein, the preset first average difference is smaller than the preset second average difference; The steps to update the analysis model include: If the average byte difference of the output result of the analysis model is greater than the preset first average difference and less than or equal to the preset second average difference, it is preliminarily determined that the accuracy of the analysis model update does not meet the requirements; Obtaining output results of two identical input quantities of the analysis model before and after the update, respectively, and calculating the difference in byte quantities of the output results of the analysis model; Comparing the difference in the byte amount of the output result of the analysis model with a preset first difference amount and a preset second difference amount respectively; If the difference in the byte amount of the output result of the analysis model is greater than the preset second difference, it is secondarily determined that the accuracy of the analysis model update does not meet the requirements, and the analysis model is updated according to the update method of the analysis model.

2. The intelligent analysis method based on industrial data visualization according to claim 1 is characterized in that: The corresponding number of repeated training times is determined by the difference between the average byte difference of the output result of the analysis model and the preset second average difference.

3. The intelligent analysis method based on industrial data visualization according to claim 2 is characterized in that: The updating method of the analysis model is: when the output result of the analysis model is a data set, the maximum number of update types of the analysis model is configured as a first update number; When the output result of the analysis model is a single data, the maximum number of update types of the analysis model is configured as the second update number; The first update quantity is greater than the second update quantity.

4. The intelligent analysis method based on industrial data visualization according to claim 3 is characterized in that: The step of determining the first corresponding backup frequency includes: If the difference in bytes of the output results of the analysis model is greater than the preset first difference and less than or equal to the preset second difference, it is preliminarily determined that the comprehensiveness of the regression test does not meet the requirements; Obtain the number of the same error types in the analysis model for calculation; Comparing the number of the same error type occurring in the analysis model with a preset number; If the number of the same error type appearing in the analysis model is greater than the preset number, it is determined that the comprehensiveness of the regression test does not meet the requirements, and the first corresponding backup frequency is determined.

5. The intelligent analysis method based on industrial data visualization according to claim 4 is characterized in that: The first corresponding backup frequency is determined by the difference between the number of the same error type occurring in the analysis model and the preset number.

6. The intelligent analysis method based on industrial data visualization according to claim 5 is characterized in that: The step of determining the second corresponding backup frequency includes: Comparing the response delay time of the analysis model with a preset delay time; If the response delay time of the analysis model is greater than the preset delay time, it is determined that the real-time performance of the data update does not meet the requirement, and the second corresponding backup frequency is determined.

7. The intelligent analysis method based on industrial data visualization according to claim 6 is characterized in that: The second corresponding backup frequency is determined by the difference between the response delay time of the analysis model and the preset delay time.

8. An intelligent analysis system using the intelligent analysis method based on industrial data visualization as described in any one of claims 1 to 7, characterized in that: include; A data collection module is used to collect some basic industry data to form a data set; A data preprocessing module, connected to the data acquisition module, for preprocessing the data set to output a control data set; A model training module, which is connected to the data acquisition module and the data preprocessing module respectively, and is used to perform feature analysis on the industry feature data in the reference data set to determine the analysis model to be trained, and to train the analysis model to be trained using the reference data set to output an analysis model; A storage module, which is connected to the data acquisition module, the data preprocessing module and the model training module, respectively, and is used to store the training information and update information of the data set, the control data set and the analysis model, respectively; A control module, which is respectively connected to the data acquisition module, the data preprocessing module, the storage module and the model training module, and is used to determine the number of repeated training times of the data training process according to the average byte difference of the output results of the analysis model, update the analysis model according to the average byte difference of the output results of the analysis model and the difference in the byte amount of the output results of the analysis model, and determine the first corresponding backup frequency of the analysis model update data according to the difference in the byte amount of the output results of the analysis model and the number of the same error types occurring in the analysis model, and determine the second corresponding backup frequency of the analysis model update data according to the response delay duration of the analysis model.

Images